Security audit
Security checks for vulnerabilities and agentic risk
This skill can directly drop database indexes based on a local recommendation file, while its headline description understates that destructive behavior.
Review this carefully before installing. Only use it on a database where index deletion from /root/.schema/recommendation.json is intended, reviewed, backed up, and safe to run; do not treat it as a general migration-file runner without adding confirmation and rollback controls.
The skill’s declared purpose is applying pending migration files, but the documented behavior actually reads an external recommendation file and performs destructive index deletion. This mismatch can mislead operators and automation into granting broader trust or running the skill in inappropriate contexts, increasing the chance of unintended schema changes against production databases.
The documented behavior is not general migration application but targeted index cleanup based on recommendation JSON, which is a materially different and more specialized destructive action. This can cause unsafe invocation, especially in CI/CD or automated database workflows that expect idempotent, reviewed migration files rather than externally generated recommendations.
The skill metadata says it applies pending migration files, but the implementation instead reads a separate recommendation JSON and performs destructive schema changes by dropping indexes. This mismatch is dangerous because it conceals privileged database-modifying behavior behind a benign migration description, increasing the chance that operators run it without understanding that it can remove performance-critical or integrity-relevant indexes.
The front matter advertises a migration runner, while the body describes recommendation-driven index deletion. In a skill ecosystem, this inconsistency is security-relevant because reviewers, orchestrators, or users may approve execution based on the safer-seeming manifest and miss that the skill performs destructive schema operations from an external input file.
The documentation describes dropping indexes but omits warnings about operational impact such as degraded query performance, locking, failed uniqueness guarantees if the named index enforces constraints, and production instability. In database administration contexts, destructive schema changes without explicit warnings or safeguards materially increase the risk of accidental service impact.
The code drops indexes directly based on external JSON input with no user-facing confirmation, dry-run, or policy gate before committing each change. In this skill context, that is more dangerous because the tool presents itself as a migration runner, so an operator may reasonably expect controlled schema updates rather than immediate destructive removal of database indexes from an auxiliary file.
The natural-language instructions and usage guidance are written in Chinese, while the file does not indicate that the skill is region-specific or provide an opt-in language choice. This can violate language/locale policy when a specific language is imposed without user consent or documented justification.
